alphabet is "worth" 2.45 trillion on the public market, is anthropic worth a bit less than 10% of google going forward? I don't think that's entirely unreasonable...
I thought the same when choosing to invest in Intel rather than NVidia in 2022. At the time, Intel was worth $310B while NVidia was worth $650B, yet Intel's revenue was $80B/year while NVidia's was $25B. I was like "There's no way I'm paying 2x the price for 1/3 the revenue." Now, NVidia is worth $4T (a return of roughly 7x) on revenue of $165B, and Intel is worth $105B (a return of roughly -66%) on revenue of $53B.
Investors are forward looking, and market conditions can change abruptly. If Anthropic actually displaces Google, it's amazingly cheap at 10% of Alphabet's market cap. (Ironically, I even knew that NVidia was displacing Intel at the time I invested, but figured that the magnitude of the transition couldn't possibly be worth the price differential. News flash: companies can go to zero, and be completely replaced by others, and when that happens their market caps just swap.)
Investors are forward looking, except when it's micron in 2000.
Anthropic have several similiar competitors with actual real distribution and tech. Ones that can go 10x are underdogs like Google before IPO or Amazon, or Shopify etc. Anthropic current stock is beyond that. Investors no longer give any big opp. to public. They gain it via private funding
So all it takes is Anthropic 35x-ing their revenue once they start selling ad spots? That sounds pretty reasonable to me.
Right now nobody wants to be the first to offer advertising in LLM services, but LLM conversation history provides a wealth of data for ad targeting. And in more permissive jurisdictions you can have the LLM deliver ads organically in the conversation or just shift the opinions and biases of the model through a short mention in the system message
No, all it takes is Anthropic 35x-ing their revenue while Alphabet revenue somehow stays the same despite Alphabet already having a product perfectly competitive with Anthropic and which can use the same revenue growth strategy.
As I said, insane. And that’s not even considering the 10 to 15% shares of Anthropic actually owned by Alphabet.
Anthropic valuation is 10% of Google. The 35 to get equivalent multiple is correct (well actually closer to 7 as another comment thread rightfully pointed that Anthropic is apparently on track to multiply their revenue by 5 in 2025).
Tech Companies are valued at a multiple of next 12 months revenue, not last 12 months revenue. Since anthropic grew from $1billion to $5billion in revenue in ~8 months, that means it ~10x'ed revenue y/y off of 1 billion base. If you assume even 60% of that growth is retained (low for traditional saas businesses, but who knows), then anthropic is ~10% of google in terms of revenue in mid ~2027.
Basically, 5x-ing revenue in 8 months off of a billion dollars starting revenue is insane. Growing this quickly at this scale breaks every traditional valuation metric.
> The company said its run-rate revenue has increased from around $1 billion at the beginning of 2025, to more than $5 billion in August.
So 10% of valuation for 1.5% of revenue, which grew 5x in last 6 months. Doesn't seem as unrealistic as you put it, if it has good gross margin which some expects to be 60%.
Also Google was valued at $350B when it had $5B revenue.[1]
Someone mentioned their projected ARR for 2025 is 9b. Which makes sense intuitively looking at how much I spent with them this year. So the valuation looks a bit more sane with those numbers.
Anthropic competes solely in one of Alphabet multiple markets and that’s a market where Google already has a compelling competitive offer. This valuation gap doesn’t make any sense to me.
It's both insane and not unreasonable. If Anthropic's internal version of Claude Code gets so good that they can recreate all of google's products quickly there's no moat anymore.
If AI is winner take all, then the value is effectively infinite. Obviously insane, but maybe it's winner take most?
It's the techno-hubristic version of Pascal's wager. The reward for the existence of God is infinite, so it worth investing all the money in the world to create one.
> " If Anthropic's internal version of Claude Code gets so good that they can recreate all of google's products quickly"
I know you aren't asserting this but rather just putting the argument out there, but to me at least it's interesting comparing a company that has vendor lock-in and monopoly or duopoly status in various markets vs one that doesn't.
I'd argue that Google's products themselves haven't been their moat for decades -- their moat is "default search engine status" in the tiny number of Browsers That Matter (Arguably just Chrome and Mobile Safari), being entrenched as the main display ad network, duopoly status as an OS vendor (Android), and monopoly status on OS vendor for low-end education laptops (ChromeOS). If somehow those were all suddenly eliminated, I think Google would be orders of magnitude less valuable.
Is there no moat for previous account and user buy-in?
Convincing billions of users to make a new account and do all their e-mail on a new domain? A new YouTube channel with all new subscribers? Migrate all their google drive and AdSense accounts to another company, etc?
It feels a bit unreasonable to me. Anthropic is arguably comparable to Google's Gemini program. Is Gemini 10% of Alphabet's value? If so, how much of that is because of its ability to consume and interact with things like YouTube and Workspaces?
I could see two or three percent, but this seems like a pretty big stretch. Then again, I'm not a VC.
Don't those cost like $400,000 a piece to outfit, though? I mean this with tremendous respect because I think they're the only ones doing it "right," I feel like Waymo is kind of 'bruteforcing' autonomous driving using money. There's an inherent limit to the impact of a technology (and thus its long-term value) based on its cost, and even stipulating that Waymo has solved it in general, I think a valuation should be contingent on a roadmap which shows how it's going to scale out -- this seems like an as-yet unsolved problem until someone shows how to combine the reliability of the tech-heavy Waymo system with the price tag of a Tesla.
Historically speaking there was an 80 year period in which transporting mined, natural, lake ice from the US Northeast/Norway around the world was economically competitive with ice machines depending on local market conditions.
Machine ice became competitive in India and Australia in the 1850s, but it took until the start of World War 1 (1914) for artificial ice production to surpass natural in America. And the industry only disappeared when every household could buy a refrigerator.
Self-driving doesn't have to scale globally to be economically viable as a technology. It could already be viable at $400k in HCOL areas with perfect weather (i.e. California, Austin, and other places they operate).
One of the most interesting statistics about Waymo is how few of them there are. The only service area with what you could call a large number of vehicles is the Bay Area. The news reports I've seen about it say under 1000 there and fewer than 3000 nationally. Uber's CEO was quoted as saying that a Waymo completes more rides than 99% of Uber drivers. It's a pity he didn't make a comparison against the median Uber driver. But it's plausible that a Waymo could replace 10 Uber drivers or more. That ratio flows through to revenue.
>Rust has effectively made writing with the performance of C++ feel like writing Ruby, but with unparalleled low defect rates and safety on account of the type system.
This is a little overblown.. speaking VERY HAND-WAVILY, sea_orm < active record by a factor of about 10x more mental overhead but is at least that much more performant...
but yea, vibe-coding rust micro services is pretty amazing lately, almost no interactions with borrow checker, and I'm even using cucumber specs...
I currently wouldn't recommend any Rust ORM, Diesel included. They're just not quite ready for prime time.
If you're not one to shy away from raw SQL, then SQLx is rock-solid. I actually prefer it over ORMs in general. It's type-checked at runtime or compile time against your schema, no matter how complex the query gets. It manages to do this with an incredibly simple design.
It's like an even nicer version of Java's popular jOOQ framework, which I always felt was incredibly ugly.
SQLx might be my very favorite SQL library of any language.
We are either limited by compute, available training data, or algorithms. You seem to believe we are limited by compute. I've seen other people argue that we are limited by training data. It is my totally inexpert belief that we are substantially limited by algorithms at this point.
I think algorithms is a unique limit because it changes how much data or compute you need. For instance, we probably have the algorithms we need to brute force solving more problems today, but they require infeasible compute or data. We can almost
certainly train a new 10T parameter mixture of experts that continues to make progress in benchmarks, but it will cost so much to train and be completely undeployable with today’s chips, data, and algorithms.
So I think the truth is likely we are both compute limited and we need better algorithms.
There are a few "hints" that suggest to me algorithms will bear a lot more fruit than compute (in terms of flops):
1) there already exist very efficient algorithms for rigorous problems that LLMs perform terribly at!
2) learning is too slow and is largely offline
3) "llms aren't world models"
General intelligence exists in this world, the inability to transfer it to a machine does seem like an algorithm problem. When it’s here we don’t even know if it will be an llm, no one knows the computer requirements.
We are limited by both compute and available training data.
If all we wanted was to train bigger and bigger models we have more than enough compute to last us for years.
Where we lack compute is in scaling the AI to consumers. Current models take too much power and specialized hardware to be be profitable.
If AI was able to improve your productivity by 20-30% percent but it costed you even 10% of your monthly salary, none would use it. I have used up $10 worth of credits using claude code in an hour multiple times. Assuming I use it continuously for 8 hours every day in a month, 10 * 8 * 24 = $1920. So its not that far off the current costs or running the models.
If the size of the models scales faster than the speed of the inference hardware, the problem is only going to get worse.
I too believe that we will eventually discover an algorithm that gives us AGI. The problem is that we cannot will a breakthrough. We can make one more likely by investing more and more into AI but breakthroughs and research in general by their nature are unpredictable.
I think investing in new individual ideas is very important and gives us lot of good returns. Investing in a field in general hoping to see a breakthrough is a fool's errand in my opinion.
If the LLM is multimodal would more video and images improve the quality of the textual output? There’s a ton of that and it’s always easy to get more.
The claim that "any amount may have negative effects" (in the article) is NOT the same as the claim that "you can detect if I had a single drop 20 years ago" (the claim you were arguing against.)
> I’m saying that in actuality, you won’t be able to see any damage if somebody drank a beer once, because there isn’t any.
Which, again, is not a claim that was ever made in the article. You are literally making things up and arguing against your own made up arguments, and framing it as if the article originally made them. No where in the article does it say that.
It's also weird that you would have a such a vehement reaction towards the quote
> "recent research indicating that any level of alcohol consumption may negatively affect health"
to the point of making things up. How on earth is such a statement even at all remotely controversial?
>VPC peering used to be annoying; now there are better options like Transit Gateway, VPC sharing between accounts, resource sharing between accounts, and Cloud WAN.
Idk, I just read the Blake Ross' (of Firefox) article [1] on having aphantasia, and identify less, rather than more, with how he describes it:
"1. Can you picture my face? >No. But it’s not personal.
2. So you don’t know what I look like? >I know facts ... If you have radiant blue eyes, I may have stored that information. ... I’m unable to project it visually in my mind because there’s no screen.
4. How about picturing something simpler, like a red triangle, or the table right in front of you? >I can’t even understand the question. I can think about the idea of a red triangle. But it’s blackness behind my eyes. Blackness next to my ears. Blackness in every nook and kindle of my brain.
5. You’re just assuming that others can actually SEE things with their eyes. NOBODY can do that, you hypochondriac.
I get it. It’s a “mind’s eye.” I don’t have it."
It's unconvincing. They sort people into categories by a questionnaire, and then find they perform differently on tests that have something to do with vision or imagination. That's it, that's all they've come up with.
Yeah, it was way above my paygrade. But I'd speculate the video's p'n'p was in the low-$xx,xxx range.
I remember when a different startup (that began inside a barn) moved from its first few h.sqft location into a proper industrial-zoned k.sqft facility. I remember when that place got its first short-schoolbus-sized Haas CnC... then got its second!
Now its had hundreds of employees (I was employee #2, non-co-founder, in this memory). I have wired in several p'n'place machines there and even the current model isn't as nice as in the video's =D
reply